A paper on Wireless Sensor Networks and the Food Industry.

1. Sensor Networks and the Food Industry
Martin Connolly Fergus O’Reilly
CEO & Technical Director,
Centre for Adaptive Wireless Systems
Sykoinia Ltd
Cork Institute of Technology
Cork, Ireland
Cork, Ireland
Tel: +353-87-2224378
Tel: +353-21-4326342
martin_connolly@sykoinia.com
foreilly@cit.ie
ABSTRACT 2005 160 ill after eating in a UK restaurant. UK Retail
Over the past decade there have been a myriad of scandals and 2005 Sudan 1 Cancer-linked colorant in Food UK Food
incidents involving contamination of the Food Supply Chain Chain. Proc.
which has drained consumer confidence and has had untold 2004 Fake infant formula causes baby deaths China Food
financial, political and health costs. Despite this, much of the Proc.
processes in the Food Industry are still manual. This paper 2004 Raw meat found to be injected with UK Food
examines the role that Sensor Networks can play in providing water and additives to retain moisture. Proc.
the industry with an automated software solution that will help 2003 Fish found to be laced with mercury and USA Fishing
prevent repeats of the contamination of the food supply chain PCBs
that is occurring on a seemingly regular basis. 2002 Cheap and standard meats mixed and Japan Food
sold at premium prices. Proc.
1. THE FOOD INDUSTRY AND ITS 2002 Altered expiry dates on butter products. Japan Food
Proc.
SCANDALS
Over the past decade there have been several incidents that 2002 Wheat containing cancer-linked nitrofen Germ. Farming
have shaken consumers’ faith in and perception of the Food herbicide fed to chickens in organic
farms
Industry. In Europe, BSE is probably the most well known
scandal to affect consumer confidence. Unfortunately however, 2002 Eggs contaminated with salmonella UK Farming
this is one of many incidents that have occurred. Each incident 2002 Oysters contaminated, ‘Norwalk-like’ Ireland Fishing
has impacts in terms of health, financial and even political virus.
costs. 2002 Expired Meat Products sold to Norw. Retail
consumers.
European Union countries have suffered several high profile 2001 GM corn found in Taco Bell products. USA Food
scandals in recent years. The presence of cancer-causing Proc.
dioxins in farm animal feed led to the banning of pork, beef, 2000 Mass food poisoning caused by Japan Food
chicken and egg products by the Belgian authorities in 1999 bacteria-ridden milk produced in Proc.
[1]. This scandal cost the country’s farm industry €1 billion unhygienic conditions
and contributed to the electoral defeat of the outgoing 2000 Kellogg plant closes down, as it could USA Food
government. In 2002, an Irish Oyster Farm was forced to close not source corn guaranteed to be GM Proc.
free grain.
after consumers in Hong Kong suffered from the Norwalk-like
virus [2]. This was traced to the oysters feeding on a sewage 1999 Illegal steroid hormones found in Europe Farming
outflow containing industrial waste from a local hospital that animal feed, meat and soft drinks. Food
Proc.
had suffered an outbreak of what is commonly termed the
‘Winter Vomiting Bug’ (a loose term embracing viruses 1999 Dioxins in feed for poultry & pigs. Belg. Farming
classified in the "Norwalk-like virus" genus of the family 1991 Food poisoning outbreak from UK Retail
Caliciviridae). consumption in a Fast Food Chain
linked to kidney failure.
These incidents are not limited to food producers. Retailers 1990 BSE/CJD (‘Mad Cow Disease’) Europe Farming
and distributors have also been indicted in a number of 1985 Wine laced with antifreeze Aus. Wine
incidents. In 2002 NorgesGruppen, the leading Norwegian
food service conglomerate, attracted unwanted publicity when
it was found that 5 of its Meny supermarkets were selling meat Table 1 illustrates that the examples cited above are just a
well past the expiration date. Indeed, one of these shops was sample of the many events that have shaken consumer
expelled from the chain after health inspectors found that staff confidence.The consumer group is now much more conscious
was making forcemeat from expired products. of the product they purchase and consume and even companies
and restaurants not directly implicated by scandal have been
These and other incidents have had a critical influence upon forced to adjust to reflect the new reality of the health-
consumers. conscious consumer. Witness the recent marketing campaign
Table 1: Food Industry Scandals by McDonalds to re-brand their image as a provider of
Year Incident Loc. Industry ‘healthy’ meals.

2. Food Authorities throughout the world have responded to the 3. SMART SENSOR NETWORKS
growing concern among consumers and lack of legal The potential applications for Smart Sensor Networks is
framework for the Food Industry. The European Union has diverse with current uses ranging from vineyards to equipment
introduced much legislation in recent years regarding food maintenance. The communication potential for Smart Sensor
product traceability, the maximum permitted levels of Networks with external systems is cited in Hac[8] where
colorants and dioxins and food hygiene[3]. It should be noted Bluetooth is proposed as a possible communication standard
that many of these directives apply throughout the food chain between Sensors. Given the protocol’s suitability for smaller
i.e. growing, production, transports and distribution. electronics products such as mobile phones and digital camera
one would expect it to be a feasible option. The Zigbee
However, there is still much work to be done as evidenced by Wireless Technology has also been proposed for intra-sensor
the recent recall of products containing the carcinogenic Sudan communication.
Red 1 food colorant[4]. From the perspective of those in the
food industry there is a requirement for the timely supply of External interfaces such as those provided by Crossbow [10]
information. This information includes temperature, the enable the Sensors to communicate with electronic devices
presence of agents and airflow. This paper addresses the such as PCs, Laptops and PDAs. Once the data reaches these
potential for Sensor Networks to aid in the provision of a devices it can literally be transmitted to any network. Support
software solution. is provided for detecting this data transmission through the
Java API provided by the TinyOS open source forum [11].
While the incidents in Table 1 can be attributed to a variety of
factors including human error, carelessness, irresponsibility Sensor Networks are recognised as one of the fastest growing
and fraud, the incidents could n future be prevented or segments in the technology industry. Part of the reason for this
managed through the judicious use of wireless intelligent is their wide applicability. Good[12] illustrates the many uses
sensor technology. For those affected by the careless or of sensor networks for security purposes while Harbor
irresponsible actions of participants in the food chain the Research estimates that the market size could be US$1 billion
deployment of a sensors that can detect pathogens, bacteria and by 2009[13].
other potentially harmful agents. On the other hand, those who
act less than ethically can be policed more effectively by health
and food safety inspectors through the ad-hoc set-up of an
4. SMART SENSOR NETWORKS & THE
intelligent sensor network during site visits. FOOD INDUSTRY
Good[12] raises an interesting point in its illustration of the
2. TECHNOLOGY IN THE FOOD use of sensors for detecting biological and chemical agents.
While security is certainly a pressing issue and an obvious
INDUSTRY application for sensor networks the technology can also be put
Woodworth[5] outlines how the Food industry is generally to other uses, in particular the detection of contaminants in the
receptive to the use of information systems. Traditionally the Food Supply Chain. The 1999 Belgian crisis could certainly
principal use of Information Technology by Food Processing have been averted if technology existed for the detection of
companies has been in Enterprise Resource Planning (ERP) dioxins in the production of animal feed.
Software. However, in recent times companies have started
using mobile devices such as PDAs and wireless technology
such as 802.11 to provide solutions in areas as diverse as As is clear from a perusal of EU sources[3], the EU has
traceability and logistics. Indeed, the incidents outlined in the imposed heavy duties upon food manufacturers. One key
previous section have led to increasing interest in Product Life requirement for manufacturers is Regulation (EC) No.
Cycle Management Software which, in conjunction with RFID 178/2002 requiring manufacturers to keep detailed records of
(Radio Frequency Identification Technology), is being mooted their supply chain. Despite the fact that this legislation has
as a solution for addressing food safety [6]. been effective since the 1st January 2005 the manual processes
still practiced by many companies has made it difficult for
them to comply with this legislation.
As pointed out in Friend[7], ERP software is designed for
managing operations within the enterprise only. However,
given a typical food processor’s dependencies on external Among the requirements are labelling and barcodes for product
suppliers and the external environment this is no longer batches so that ‘problem’ batches can be easily recalled. Not
adequate for today’s food industry. Indeed, many of the key only is this good for the consumer it is also good for the
tasks in food production are still carried out manually, For producer and retailer as products can be recalled in a more
example, growers still manually measure temperature and food timely fashion. This is a key point the legislation does not
scientists often have to manually carry out tests to detect mandate the publicising of product recalls if this process
ingredient composition and check for the presence of occurs before the product enters the consumer market so
contaminants. Mobile technology is one factor in providing an companies can avoid the attendant adverse publicity and
automated software solution for the food industry but the damage to their reputation.
solutions offered today only cater for the tagging and labelling
of products. Granted, these are key components of a good However, there are issues with labels and barcodes, which can
traceability system and are vital for the post-auditing of food reduce their effectiveness. Barcodes and labels are line of sight
contaminations but they don’t prevent the actual problem. It is only and are unreadable if they are damaged or soiled. The
in addressing that intelligent sensors can provide a vital role. emerging solution to this problem is the use of Radio
Frequency Identification (RFID) tags. RFID tags are essentially
sensors that contain unique ID for a product or product batch.

3. The tags are long range and since they are a radio technology growers, for example, have very specific requirements in terms
can be embedded in plastic. Of course, since they are Sensors of light, temperature and airflow and have to manually
they can form a part of an overall Sensor Network. Roberti[14] measure these attributes. The deployment of a Smart Sensor
demonstrates how the US Navy used RFID as part of an overall Network would greatly increase the efficiency of this process.
Sensor Network to monitor pressure, temperature and humidity
in shipping containers for aircraft parts. 6. CASE STUDIES
6.1 The Wine Industry
The point raised in Roberti[14] can also be applied to the Food One of the most popular early deployments of Sensor Networks
Industry. Many products require constant monitoring is in the Wine Production Industry[15]. However, there are
throughout their supply chain. This is not only for compliance other successful examples in different parts of the world, such
purposes but also to ensure that the basic quality assurance as how Pickberry Vineyard in Sonoma, California was able to
requirements to retailers are met. For example, the chill chain manage its crop growing problems through the use of a Sensor
requires that chilled products are stored at a constant Network provided by Accenture Labs[16]. From the
temperature throughout the delivery and storage of the product. perspective of vine growing sensors can be used to monitor soil
Typically each product batch is examined by the retailer on moisture, rainfall, wind velocity and direction, and air and soil
arrival and is rejected if it does not meet the required temperature. The monitoring of these factors can be critical in
temperature. Up to now, the only solution to this problem for the cost management for a vineyard. For example, detecting
many smaller companies was to take periodic manual frost in a timely manner can prevent loss of a crop, all the
temperature readings. With the advent of Sensor Networks it is more important since, to take the Pickberry example, each ton
now possible to devise a solution whereby the producer is of grapes can be worth US$1000 to US$4000.
notified virtually instantaneously if the temperature failed
outside acceptable parameters.
However, while all the focus is currently on using Sensor
Networks for the growing of grapes this is merely one of the
5. SENSOR NETWORKS & FOOD stages in wine production for which intelligent sensors can
GROWERS play a role. For example, temperature must be strictly
Of course, food processing companies are only one segment of controlled during the vinification process (i.e. the conversion
the Food Industry. Growers also have requirements that can be of grape juice into wine). This is so that yeasts attached to the
met by sensor networks. The most well know example is that grape can feed on the sugar to produce alcohol – for white
of the ‘smart vineyard’. The Australian Cooperative Research wine the temperature must be between 15°C to 20°C for
Centre (CRC) for micro technology and Motorola have fermentation, for red wine fermentation requires a temperature
devised a Sensor Network solution for the Australian Wine of between 25°C and 30°C. Furthermore, for this controlled
Industry[15]. The devices used in this project can measure fermentation process the presence of Sulphur Dioxide must be
wind speed and direction, temperature, light, humidity, soil added but this is monitored in accordance with strict European
moisture and leaf wetness. Given that weather conditions and Union guidelines.
the anticipation of same is a key factor in the production of
quality wine it is obvious that the use of a Sensor Network can
Intelligent Sensors can
be of great benefit to vine growers.
clearly play a role in the
fermentation process whether
Such a system also has the potential to give vine growers better it is in temperature or
information for monitoring the crop and can ultimately play a Sulphur Dioxide monitoring.
role in anticipating problems such as the presence of pests However, these two
such as phylloxera, an insect that sucks fluid from the grape examples are just one of the
vine which results in the rotting of the plant. many potential uses for
Sensor Networks in
The ubiquitous nature of the radio transmission of Sensor vinification – intelligent
Networks is also important here. By its nature agricultural land sensors play a role in
can vary widely and the same applies to vineyards. The everything from detecting the
Sensors can be deployed virtually anywhere in a vineyard as presence of malic acid to
long as they are in requisite range of each other. The other key tannins.
advantage here is cost. Traditionally climate sensors are The final step in the production process of course is storage.
expensive which precludes their use from most growers and Once again sensors can be used here to detect levels of
even then only on a limited scale. humidity and temperature in a cellar. Randomly deployed
sensors could also be potentially be used to detect what is
The above characteristics of sensor networks are not just useful commonly termed ‘corkage’ (air contact with the wine due to
to vine growers. Any fruit grower will have similar inadequate corking) in wine bottles or batches.
requirements, as indeed will growers of cereals, vegetables and
sugar beet. Once again the potential use of Sensor Networks is
wide ranging. 6.2 The Chill Chain
One of the greatest challenges for the Food Industry is for the
Monitoring of crops is at present performed manually and on a manufacturers of frozen foods, in particular meat and poultry.
somewhat ad-hoc basis. At times, this is a very onerous task Temperature must be maintained at a constant level from
given the very specific requirements of a crop. Mushrooms initial processing to final display by a retailer. However, given
the steps from production to sale this has up to now been an

4. onerous though essential requirement for frozen goods for analysing the data. The system will literally have to work
producers. This term coined for this issue is the Chill Chain. from ‘out of the box’. While this is admittedly a difficult goal
for any emerging technology it is critical to the success of
As implied above the Chill Chain consists of a number of Sensor Network Systems in the Food Industry.
steps. Primary chilling relates to removing the heat from the
carcass before it can be further processed or shipped. However, However, while deployment is a challenge, smart sensors
there can be startling differences in the surface temperature of already monitor the actual measurements required. Critical
a carcass and its deep temperature[17]. Once a previously measurements such a temperature, wind speed and humidity
chilled produced has been cut, minced, wrapped or cooked are already catered for and there is little that is new. One open
secondary chilling must take place. This is vital for ensuring research topic is the development of Sensors that can measure
that a product remains at a constant temperature during not only the presence of a particular ingredient or characteristic
transportation. but also the quantity or percentage of that characteristic. For
example, it would be useful not only to whether there is
Studies have shown that primary and secondary chilling are limestone in the soil but also the percentage composition of the
vital for maintaining potential shelf life. In studies[17] it was soil that is limestone. Similarly, while the presence of dioxins
found that vacuum packaged beef on average achieved only can be tested with current sensor technology the level of
25% of its potential shelf life, 2 weeks instead of 8 weeks. dioxins in products such as milk cannot currently be tested
This is due to a number of factors including loading the meat using smart sensors. Such information is important for quality
for transport at too high a temperature and the incomplete assurance programs and has to be carried out manually for the
cooling of boxed meat due to weak air movement. most part.
The other point of interest regarding the Food Industry is its
There is an obvious opportunity for sensor networks in primary use of relatively few and relatively generic software systems.
and secondary chilling, not only to monitor temperature but The output from any Sensor Network System must be easily
also to monitor airflow. Currently, despite the best intentions transferred to third party software such as MS Excel and SAP.
of manufacturers the possibility of error is very high. Not only Thus, any system must be open and have easy to use or even
is this a risk for the consumer but there is also much wastage if automated export functionality to other software.
shelf life is being reduced to a quarter of its potential.
8. FAILINGS OF SENSOR NETWORKS
Primary and secondary chilling take FOR THE FOOD INDUSTRY
place in plant but these are only two
From the authors’ experience, there are a number of factors
steps in the Chain. Refrigeration
that act as serious barriers to entry for the use of Sensor
when transporting the goods much
Networks in the Food Industry:
provide an adequate air distribution.
It should be noted that the
refrigeration unit does not cool the • Reliability
goods. Rather it maintains them at the A recent deployment of Sensor Networks in two redwood
temperature at which they were trees in Sonoma, California found that 65% of the nodes
loaded. There are similar never returned data [18]. Similar problems were recorded
requirements for Off-site Cold when using Sensor Networks for habitat monitoring[19].
Storages, Central Distribution The former statistic would be unacceptable in any
Centres and Wholesalers and, of commercially deployed food monitoring network and it
course, Retailers. would be very difficult to make a business case for a
network where a significant proportion of the nodes never
work.
While primary and secondary chilling are the responsibility of
the processor, essentially manufacturers have no control once • Ease of Use
the goods leave their premises particularly if they do not have Food processing organisations tend to be technology
their only transportation. Ultimately, any fault in the chill agnostic. Microsoft based technology prevails as it is easy
chain will damage a frozen goods manufacturer’s reputation to use and install. For the non-technical user Sensor motes
even if this fault is caused by a third party. However are intimidating to deploy and use. Indeed, implementing a
mandating the use of a sensor network to monitor temperature Sensor Network lies outside the range of the average Food
and air flow in the other steps in the process can play a critical Industry Information Systems Professional.
role in the monitoring of food quality and can help prevent • Data gathering and Data Interpretation
spoiled goods reaching the end consumer. While Sensor Networks undoubtedly can play a significant
role in improving the availability of update data on a
7. FOOD INDUSTRY SYSTEM physical environment there is currently no standard way for
REQUIREMENTS gathering and interpreting the data. The Food Industry
The Food Industry involves complex processes and detailed relies heavily on standard reporting tools such as Crystal
regulation. By contrast, the typical Information Systems Reports and Brio and would be loathe to use non-standard
department wants software that is easy to use [5]. Any Sensor Java-based reporting GUIs.
Network System that is targeted for the Food Industry will • Organising Networks and Clustering
have to be easy to deploy. Food Processors are uninterested in
configuring sensors, their interfaces and the attendant software

5. There is at present no mechanism by which end users can Fergus O’Reilly is a lecturer in the Department of Electronic
organise their sensors for reporting purposes. Granted, Engineering in Cork Institute of Technology and a Principle
group IDs are available but there can be limited – for Investigator in the Centre for Adaptive Wireless Systems, Cork
example, two sensor groups might be monitoring the same Institute of Technology. (http://www.aws.cit.ie)
area and would be effectively the same sensor group from a
reporting perspective. What is required is a means by 12. REFERENCES
which users can organise their groups into clusters that [1] van Larebeke, Nik, Hens, Luc, Schepens, Paul, Covaci,
make sense to them. Adrian, Baeyens, Jan, Everaert, Kim, Bernheim, Jan L,
Vlietinck, Robert, and De Poorter, Geert, “The Belgian
PCB and Dioxin Incident of January-June 1999: Exposure
Data and Potential Impact on Health, Environmental
Health Perspectives Volume 109, Number 3 March 2001
[2] Cassidy, Eddie, “Norwalk-like Virus Outbreak in Hong
Kong Forces Closure of Irish Oyster Farm”, The Irish
Examiner 26/03/2002
[3] Official Journal of the European Union
[4] Food Safety Authority of Ireland, FSAI Alert Notifications,
http://www.fsai.ie/alerts/fa/index.asp
[5] Woodworth, Simon. “Track and Trace: An opportunity for
Information Systems Deployment and Supply Chain
integration”, Food Traceability Seminar, NUI Cork,
February 2005
[6] Sullivan, Laurie, “PLM Software Has A Role In Food
Safety”, InformationWeek, Jan. 12, 2004
Figure 1: Sensure Sensor Portal [7] Friend, Bill and Thompson, Olin, “Managing Software: Is
the food industry ready for Product Life Cycle
9. THE SYKOINIA SOLUTION Management?”, Food Industry Magazine March 2003
The above issues effectively preclude any commercial use of [8] Hac, Anna, “Wireless Sensor Network Designs”,Wiley
Sensor Networks in the Food Industry. Sykoinia’s SenSure Press 2003
solution aims to bridge this significant gap in Sensor Network
technology, by providing a means for end users to organise [9] The ZigBee Alliance http://www.zigbee.org
their networks, providing integrated Crystal Reports [10] Crossbow Technology Inc. http://www.xbow.com
functionality and enabling users to add different types of
sensors easily and quickly. [11] Tiny OS http://www.tinyos.net/
[12] Good, Barbara G, “Sensors for Security”, Sensors
Data produced by the sensors is collated by the SenSure portal Magazine, July 2004
Figure 1, and is then forwarded to the server system for
analysis and storage. The server system can be accessed [13] Harbor Research, Inc http://www.harborresearch.com/
through a user interface that can organise the Sensor Groups [14] Roberti, Mark, “Navy Revs Up RFID Sensors”, RFID
and can also request and schedule reports. Journal June 2004
10. CONCLUSION [15] “MEMS Come to Oz Wine Industry”, Electronic News
Sensor Networks have a myriad of potential uses. The Food June 2004
Industry provides ample opportunities for both hardware and [16] “Accenture Prototype helps Pickberry Vineyard improve
software providers given its need for monitoring software – a Crop Management”,
niche that a sensor network consisting of diverse smart sensors http://www.accenture.com/xdoc/en/services/technology/c
can fulfil. ase/pickberry.pdf
However there are a number of caveats – a complex and hard [17] Managing your Chill Chain--The Undiscovered Country,
to use system will not succeed in this industry nor will a Meat & Poultry, February 1999
‘closed’ system where data is difficult to export, interpret and
analyse. [18] Tolle, Gilman & Culler, David, Design of an Application-
Cooperative Management System for Wireless Sensor
11. About the Authors Networks, 2ndEuropean Workshop on Wireless Sensor
Martin Connolly is the CEO and Technical Director of Networks (EWSN) Istanbul, Jan 31-Feb2, 2005.
Sykoinia, a provider of integrated Information Systems
solutions for Sensor Networks. Sykoinia’s solution has been [19] Szewczyk, Robert et al., “Lessons from a Sensor Network
deployed in orchards, vineyards and food processing halls. For Expedition”, 1st European Workshop on Wireless Sensor
further information visit http://www.sykoinia.com. Networks, Berlin, Jan 19-21, 2004.